Crashing Memory 2.0: False Memories in Adults for an Upsetting Childhood Event

UC Irvine UC Irvine Previously Published Works

Title Crashing Memory 2.0: False Memories in Adults for an Upsetting Childhood Event

Permalink https://escholarship.org/uc/item/6vx8w81s

Journal Applied Cognitive Psychology, 30(1)

ISSN 0888-4080

Authors Patihis, L Loftus, EF

Publication Date 2015

DOI 10.1002/acp.3165

Peer reviewed

eScholarship.org Powered by the California Digital Library University of California Applied Cognitive Psychology, Appl. Cognit. Psychol. 30:41–50 (2016) Published online 15 September 2015 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/acp.3165

Crashing Memory 2.0: False Memories in Adults for an Upsetting Childhood Event

LAWRENCE PATIHIS1* and ELIZABETH F. LOFTUS2 1University of Southern Mississippi, Hattiesburg, USA 2University of California, Irvine, USA

Summary: Previous crashing memory studies have shown that adults can be led to believe they witnessed video footage of news events for which no video footage actually exists. The current study is the ﬁrst to investigate adults’ tendency to report memories of viewing footage that took place when they were children: the plane crash in Pennsylvania on 11 September 2001. We found that in a computer questionnaire, 33% indicated a false memory with at least one false detail. In a more detailed face-to-face interview, only 13% of the group described a detailed false memory. Familiarity with the news story, fantasy proneness, alcohol use, and frequency of negative emotions after 9/11 were all associated with a Persistent False Memory. Participants who had received prior suggestion were more likely to later report false memories in the subsequent interview. We discuss our novel results and the importance of the paradigm. Copyright © 2015 John Wiley & Sons, Ltd.

Memory distortion research has important applications in have focused on events that happened when the participants real-life cases in the law, clinical psychology, and other were adults. In the present study, we explore the phenome- areas. Some of these real-life cases have involved the distor- non by asking adults about an event that occurred in their tion of memory in children, in adults for recent events, and in childhood. We also investigate a number of other factors that adults for events in their childhood. To investigate both have not previously been examined in crashing memory applied and theoretical issues related to such real-life phe- studies, such as the effect of prior suggestion. nomena, researchers have developed a number of memory distortion paradigms, including the misinformation effect Previous crashing memory studies (Loftus, Miller, & Burns, 1978), associative word lists (Roediger & McDermott, 1995), rich false memory (Loftus In the first crashing memory study, Crombag et al. (1996) & Pickrell, 1995), and imagination inflation (Garry, told Dutch participants there was videotape of a widely Manning, Loftus, & Sherman, 1996). One paradigm, the reported Boeing 747 crash into apartments in Amsterdam. ‘crashing memory’ technique (Crombag, Wagenaar, & van Although the actual plane crash had not been filmed, a Koppen, 1996), involves asking participants if they have remarkable 55% in Study 1 and 66% in Study 2 of the parti- seen footage for a widely reported news event, often invol- cipants reported seeing the footage and about 45% reported ving a crash (hence the name), when in fact no video footage details of its contents. A follow-up study demonstrated a of the event really exists. In response to such questions, similar phenomenon for a nationally important car crash. many participants in these studies appeared to develop false Ost, Vrij, Costall, and Bull (2002) asked participants if they beliefs and memories that they witnessed events they could had seen footage of the car crash in which Princess Diana not possibly have seen. was killed (when in fact none exists), and 44% of the sample The crashing memory paradigm has several advantages reported they had. that represented an important alternative way to measure The studies that followed reinforced these surprisingly memory distortion. The first advantage is that the event large percentages with different target news events and also involved a nationally important news story that was usually showed that social influence could boost false memory rates. personally important to the participants, thus allowing for For example, Granhag, Stromwall, and Billings (2003) the measurement of an upsetting memory that can be auto- found that 55% of respondents reported that they saw nonex- biographical in nature. The second advantage is the national istent footage of a well-known incident involving a sinking news event in question is somewhat of a collective experi- ferry and found that about 30% gave a false detail, suggest- ence across participants, and thus, their memory reports are ing that the false beliefs may have been accompanied by comparable across subjects. This is often not true in other memories. The results also showed that misleading com- studies where people are asked about their upsetting or ments from peers, when overheard by the participant, can traumatic memories, and different participants point to boost false memory rates (see also Ost, Hogbin, & Granhag, completely different types of events in their lives. The third 2006 for a replication using a different target news event). advantage is that the researcher can be reasonably confident Similarly, Wilson and French (2006) asked participants to that the event (viewing disturbing footage in this case) did recall the details of a bombing that had occurred in a Bali not take place. This certainty about the falsity of the memory nightclub, and 36% of participants reported seeing nonexis- is not always present in other autobiographical false memory tent footage, with nearly all of them reporting corresponding research (e.g., Loftus & Pickrell, 1995; Garry et al., 1996). memories for details they could not have seen. Despite these advantages, previous crashing memory studies These studies raised the question whether these false memory reports were being caused by the suggestion or whether they were spontaneously generated. An interesting *Correspondence to: Lawrence Patihis, Department of Psychology, Univer- sity of Southern Mississippi, Hattiesburg, MS 39406, USA. experimental approach by Smeets et al. (2006) found that E-mail: [email protected] the level of suggestiveness in the crashing memory questions

Copyright © 2015 John Wiley & Sons, Ltd. 42 L. Patihis and E. F. Loftus affected the rates at which participants reported seeing non- (Ost et al., 2002; Ost et al., 2006; Granhag et al., 2003; existent news footage of the assassination of Dutch politician Smeets et al., 2009). With regard to personality-related mea- Pim Fortuyn (a famous figure well known to the Dutch par- sures, Ost et al. (2008) found that those scoring high on dis- ticipants). Interestingly, even with no suggestion (‘Do you sociation (Dissociative Experiences Scale, DES-C; Wright & remember whether there was a film…’; notice the indefinite Loftus, 1999) or fantasy proneness (Creative Experiences article ‘a’), 27% of participants indicated they had seen the Questionnaire; Merckelbach, Horselenberg, & Muris, 2001) footage of the assassination, although only 6% reported were more likely to give definitive details of nonexistent actual false details. With false suggestion (‘Did you see the footage (of the bus moving in the London bus bombings). amateur film of the Fortuyn shooting?’; notice the definitive These measures relate to an ongoing debate as to whether article ‘the’), the false reports were much higher: 63% with the most prone people who develop false memories are those 33% giving false details. These results demonstrate not only that are highly hypnotizable (related to fantasy proneness), a the importance of suggestive wording but also that false question that has historical roots in false memory production memories can occur even in the absence of misleading in hypnosis (see Patihis & Younes Burton, in press). The post-event information (spontaneously; cf. Mazzoni, 2002). question of whether dissociation is related to false memories Further research established the importance of familiarity is routed on an ongoing debate about the relationship, if any, of the news event to the participant. Ost, Granhag, Udell, between dissociation and memory (e.g., Lynn et al., 2014). and Hjelmsäter (2008) asked participants, 150 from Sweden Another area of interest is the possibility that participants’ and 150 from the UK to complete questionnaires about the tendency towards social desirability might explain the explosion of the number 30 bus in Tavistock Square, unusually high frequency of false memory reports in previ- London. UK participants were more likely to say they had ous crashing memory research (cf. McCloskey & Zaragoza, seen nonexistent computer-generated image of the explo- 1985). And finally, an individual difference that may also sion, and nonexistent television footage of the explosion, affect memory is alcohol consumption, which has shown compared with the Swedish participants (40% vs. 16%). This promise as a factor in other areas of memory distortion indicated that perhaps familiarity with the event, or indeed research (e.g., Garfinkel, Dienes, & Duka, 2006), but there the amount of exposure to the original news reports, might is a scarcity of data addressing alcohol in crashing memory facilitate the implanting of such memories. studies. In the current study, we investigate the role of these A study by Smeets, Telgen, Ost, Jelicic, and Merckelbach variables and others. (2009) perhaps created doubts about how many of the false memory reports in previous crashing memory studies were The current study in fact valid false memories that would persist. In keeping with a number of past studies, they found high rates of false Although previous research seemed to establish a strong report of seeing nonexistent footage (66%) following the effect of suggestion on memory of nonexistent footage, suggestion that there is footage of the assassination of Pim Smeets et al. (2009) raised the question as to how many of Fortuyn. However, after the debriefing, in which they were these are genuine and indeed persistent false memories. They informed that there is no footage of the actual moment of also discussed an apparent reduction in false memories and the assassination, most claimed they had not fully under- the reasons for it. A key question raised by this was whether stood the question, with only 10% maintaining that they false memories in crashing memory studies were spontane- had truly remembered the footage during the experiment. ously produced by prior exposure to media or whether they This decrease in false memory percentages in a crashing were caused by the suggestion in the actual experiments. memory study raised questions whether previous studies The current study explores these issues. Previous crashing could have had inflated percentages and also what in partic- memory studies have also not examined events that took ular could have caused the reduction in percentages. In place in the childhood of the participants. The current study Smeets et al., it is unclear whether the true post-event infor- does this to gain an insight to the real-life parallel of false mation (i.e., ‘there is no footage’) caused the decrease in memories induced into adults for upsetting childhood events. false memory reports or whether the open-ended and more We also investigate a number of possible correlates of these detailed, clear, and cautious approach of the post-debriefing autobiographical false memories, such as gender, fantasy interview caused the decrease. Ost et al. (2008) and Smeets proneness, and dissociation. et al. (2009) raised the issue of whether the suggestive question in crashing memory studies simply brings out a pre- METHOD existing false memory (spontaneous; formed before the study, perhaps by exposure to successive media reports) or Participants the suggestion in the experiments causes the false memory production (suggestion-dependent; cf. Mazzoni, 2002). The participants were 297 undergraduates (221 females and 78 males) who participated in the study in exchange for course Individual differences and crashing false memory credit. Of these, 48.8% are self-identified as Asian/Indian, Some previous research has been carried out investigating 21.5% as Caucasian, 16.2% as Hispanic/Latino, 7.4% as individual differences and crashing memories. In some stud- Middle Eastern, 3.4% as Hawaiian/Pacific Islander, 2.0% as ies, women were more prone to false report (Crombag et al., African-American/Black, and 0.7% other. Participants’ mean 1996; Jelicic, Smeets, Candel, van Suijdam, & Merckelbach, age was 20.1 years (SD = 3.1). The majority of our partici- 2006a; Jelicic et al., 2006b) but not in other studies pants (n = 271) were between 18 and 22 so that they were

Figure 1. A simpliﬁed representation of the study design between the ages of 8 and 12 years at the time of the event. asked about memory for their negative emotions in the For purposes of comparison, there were also a small number week following 11 September 2001. In addition to those (n = 26) of older participants, between ages 23 and 51 years. questions, participants also completed a number of measures, including an alcohol-use scale (modiﬁed from LaBrie, Design Hummer, Grant, & Lac, 2010), the Creative Experiences Questionnaire (fantasy proneness; Merckelbach, Muris, & Participants were randomly assigned to one of two condi- 1 Rassin, 1999; Merckelbach, Horselenberg, & Muris, 2001), tions : an experimental condition (in which they completed and the DES-C (Wright & Loftus, 1999). Session 1 typically a computer questionnaire containing suggestive information took participants about 35 minutes to complete. about the target event) or a control condition (in which they completed a computer questionnaire about an unrelated event). Session 2 Approximately 40 minutes after the questionnaires, all partici- Exactly 1 week after Session 1, the subjects returned to the pants underwent a face-to-face interview in which they were lab for Session 2. As in Session 1, a research assistant gave asked about their memories of the target event. See Figure 1 verbal instructions and then helped the participant start on for a graphical representation of the study design. the computer questionnaire.

Materials and procedure Crashing memory questionnaire Participants were invited to participate in a study ostensibly Depending on which condition they were assigned to, parti- about ‘Personality, Individuality, and Slideshows’. All sub- cipants either completed a news story questionnaire asking jects participated in 2011, between March and November, about their memory for United 93 crash footage (experi- approximately 10 years after 9/11. Participants came into mental condition), or the Human Genome news event the laboratory one at a time, staggered one every (control group). Each questionnaire was similar in construc- 15–20 minutes, and were greeted by a research assistant tion and length and differed only in the memory recall target who gave them verbal instructions to prepare them for the (i.e., the targets were United 93 crash footage, versus the study. Between one and three participants participated in a Human Genome news event). See the Supporting Informa- computer laboratory room at any given time, with one or tion for the crashing memory materials used in the computer two research assistants supervising. The lab room was questionnaire. windowless and mostly silent. The research assistants who interacted with the participants and conducted the interview United 93 condition were blind to both the assigned condition of participants In this condition, participants were told that footage of the and the precise hypotheses of the study. crash exists and has been widely shown and were then asked whether they had seen the footage. They were then asked to indicate details of the footage, which involved force choice Session 1 questions involving false details, including an option of ‘I Once randomly assigned to condition, the participant don’t remember’. For example, one question asked about proceeded to fill out the computer questionnaires. They the participants’ memory of the video footage and gave the first answered demographic questions. Other measures choices that the plane (a) ‘came down vertically, nose down ’ ‘ 1 and almost without forward speed , (b) slid into the ground In order to keep the current analysis focused and clear, 96 participants who ’ ‘ ’ were randomly assigned to a third condition are not analyzed in the current almost horizontally and at considerable speed , or (c) I can t study because by design they did not participate in the interview (Patihis, remember’. Another question gave the participants the 2012). The random assignment was set up on the computer to place one-half choice that the video footage was (a) ‘very clear, you can of participants into the experimental condition and one-quarter into each of ’ ‘ the two control conditions, resulting in 202 in the experimental condition see and hear exactly what is happening , (b) fuzzy, it is and 95 in the human genome control condition in the main article difficult to tell what is happening’, or (c) ‘I can’t remember’.

For the full wording of the questionnaires, see Supporting Quantitative coding of interview responses Information. Participants then proceeded to fill out a All interviews were coded by two independent coders Memory Characteristics Questionnaire (MCQ, modified from (research assistants), and any inter-rater disagreements were Johnson, Foley, Suengas, & Raye, 1988 to include an I don’t scrutinized carefully and resolved by a supervising remember option, scored as 0) about their memory for the researcher in discussion with one or both research assistants. footage. This condition acted as the experimental condition, The question asking about whether the participant was fami- simulating the crashing memory manipulations of previous liar with United 93 was coded as 0 = no, 0.5 = unsure/ studies. maybe, and 1 = yes, and the initial inter-rater agreement rate was 78% (66 disagreements out of 297; Cronbach α = .839). Control condition Whether someone was familiar with the United 93 news In this condition, neither United 93 nor 11 September 2001 event was not always easy to code, because sometimes was mentioned at all. Instead, participants were asked a participants would start out by saying ‘a bit’ but then go on similar set of questions about their memory for the Human to give details to demonstrate that they were fully familiar. Genome news event (as opposed to footage) that was These difficult-to-code cases were carefully recoded when- reported in 2001. This condition acted as a control condition ever research assistants’ initial coding did not match. The that not only had no suggestion of nonexistent footage, but questions asking whether they had seen the footage were also did not stimulate any imagery of United 93 or 9/11. coded in a similar manner (no =0;maybe/unsure = 0.5; These two conditions allowed us to measure the effect of yes = 1), and the initial inter-rater agreement rates on those prior misleading suggestion on the interview, which occurred questions were 93% and 92%, respectively (Cronbach α = approximately 40 minutes later. .955 and .935). The question asking how well the participants remember the video on a scale from 1 to 10 was Fillers straightforward to code, and the inter-rater agreement rate Subjects completed a number of filler tasks that for reasons was 99% (Cronbach α = .996). Interviews that included a of clarity and focus are not analyzed in the current study. ‘yes’ response and a memory score of more than 2 on the Participants also completed a 91-item personality scale 1–10 Likert scale were coded for false details of the actual designed to measure a total of 13 personality subscales crash by a supervising researcher (L. P.). All of these 43 (Swedish Universities Scale of Personality; Gustavsson cases examined had the coding justified by transcribing the et al., 2000). These trait subscales include somatic trait relevant part of the interview text (performed by a research anxiety, psychic trait anxiety, stress susceptibility, lack of as- assistant or the supervising researcher) and considering whether sertiveness, impulsiveness, adventure seeking, detachment, the actual content of the detail must be false (i.e., movement of social desirability, embitterment, trait irritability, mistrust, the plane or the actual impact/explosion was coded as a false verbal trait aggression, and physical trait aggression. The detail, whereas pictures or video of the aftermath were not social desirability scale, in particular, was used to investigate assumed to be false). Details from any of the other crashes on the participants’ wish to help the researchers (i.e., demand 9/11 (e.g., a crash into a building) were not coded as a false characteristics) as a possible confound. In addition, research detail. assistants who conducted the interviews also completed this personality scale as well as demographic questions (for Results reasons of space and clarity, the results for interviewer characteristics are mentioned only briefly in the main article, In the results that follow, we examine the rates of false mem- and in more detail in the Supporting Information). ory in the questionnaire and then the interview. We then examine the association between false memory and familiarity United 93 crash footage interviews with the news story, prior suggestion, ethnicity, alcohol use, In this audio-recorded structured interview, participants were emotion, and fantasy proneness. Finally, we take a specific taken away from the other participants into a quiet room. In a look at those participants who were between the ages of 8 one-on-one and face-to-face conversation with the research and 12 years at the time of the target event (9/11). assistant, participants were told about the various crashes on 9/11, then told that we are focusing here only on United Computer questionnaire 93, and asked ‘Are you familiar with this event?’ (open re- Of the 297 participants completing the study, 95 were sponse). The participants are then told that there is footage assigned to the Human Genome control no-suggestion ques- for the crash of United 93, and then asked ‘Do you remember tionnaire,2 and 202 were assigned to the crash footage ques- seeing that footage?’ (open response). No such footage actu- tionnaire (for an explanation of the unequal group sizes, see ally exists. In this interview, we took pains to be clear about Footnote 1). Of the 202 participants assigned to the United what crash we were referring to and reiterated it was a crash 93 crash footage condition, when asked ‘have you seen the into a field. See Supporting Information for the interview video?’, 36.6% indicated ‘yes’, they had seen the United script. In the interview, those participants who said ‘yes’, 93 crash footage. Of the 36.6% who indicated ‘yes’, they they had seen the footage were then asked follow-up ques- had seen the footage, 91.8% gave at least one false detail tions about details. After the interview was complete, we revealed that the study was actually about memory distortions 2 By definition, the Human Genome condition (n = 95) reveals nothing re- and that there is no footage of the United 93 crash. The typ- lated to false memories at the questionnaire stage; hence, you only see those ical duration of Session 2 was between 60 and 75 minutes. participants in the analysis of the interview that came 40 minutes later.

(how the plane moved, clarity of footage, or length of foot- three conditions were met in the interview: the participants age). Specifically, of those 36.6% who said ‘yes’, 62.2% says ‘yes’, they have seen the video, they give a score of 2 gave a detail about the plane moving, 60.8% gave a detail or above on the aforementioned 1–10 scale and they give about after the impact, 68.9% gave some detail about the some false detail from the footage connected to the actual clarity of the footage, and 51.4% gave a detail about the crash (most commonly being how the plane moved or length of the footage (see Supporting Information for ques- crashed in the footage). A false belief involves the partici- tion wording on the four detail questions). Of all the 202 par- pants saying ‘yes’, they had seen the footage, but then either ticipants, 41.1% gave some false detail of the nonexistent choosing 1 (no memory at all) on the 1–10 scale or giving no plane crash footage (therefore, a minority of 4.5% partici- false details of the actual crash in the footage. Using these pants initially indicated they had not seen the footage but criteria, we identified 30 (10.1%) as having an Interview consequently chose a false detail in the follow-up questions). False Memory, 18 (6.1%) having a false belief, and 249 Of the 202 participants in the group that received false (83.8%) showing neither. suggestion (the crash footage condition), when asked ‘how ’ well can you remember having seen the video? , 59.4% indi- Comparing experimental to control group cated 1 (no memory at all) on the scale from 1 to 10, and 40.3% indicated a 2 or above. A score of 5 or above was in- Effect of prior suggestion on Interview False Memory dicated by 9.9% of the participants, with one participant Some subjects who participated in the audio-recorded in- (0.5%) indicating 10 (a very clear memory). Of those 73 that terview had responded to a suggestive questionnaire about had indicated ‘yes’, they had seen the footage, all (73; 100%) United 93 on the computer 40 minutes earlier (experimen- indicated a score of 2 or above on this scale, with 26% (19) tal condition), while others had been randomly assigned indicating a score of 5 or above. In contrast, of the 128 who to a non-suggestive neutral questionnaire about the 2001 indicated ‘no’, they had not seen the video, 93.8% (120) Human Genome news story (control condition). Figure 2 indicated a score of 1 (no memory at all) on this 1–10 scale. shows that a higher proportion of those in the experimental condition exhibited Interview False Memories (13.4%; 27 False belief versus false memory of 202) compared with those in the control condition χ2 N p We classified subjects as having a Questionnaire False (3.2%, 3 of 95), (1, = 297) = 7.42, exact = .006, Cramer’s V = 0.158. This four-fold difference in false Memory if all three of the following conditions were met: they indicated ‘yes’, they had seen the footage, indicated a memory proportions in the interview shows the potent score of 2 or above on the aforementioned 1–10 scale, and effect of the suggestive questionnaire 40 minutes earlier. chose at least one false detail. A false belief involved choosing ‘yes’, they had seen the footage, but either choosing 1 (no Effect of prior suggestion on quality of interview false memory at all)onthe1–10 scale or indicating no false details memories in the follow-up questions. Using this criteria, 67 (33.2%) of Within those who demonstrated an Interview False Memory, the 202 participants indicated a Questionnaire False Memory, we examined whether there is a difference in memory quality 7 (3.5%) indicated a false belief, and 128 (63.4%) reported no between those who had prior suggestion 40 minutes earlier false belief or memory. (n = 27) and those who had not (n = 3). We found that those

Audio-recorded interview About 40 minutes after the computer questionnaire, those in the United 93 crash footage (experimental, n = 202) condition, and the Human Genome (control, n = 95) condition participated in the face-to-face recorded interview with a research assistant. Of the 297 participants who participated in the crashing memory interview, 48 (16.2%) said they had seen the footage, 23 (7.7%) said unsure/maybe, and 226 (76.1%) said they had not seen the crash footage of United 93. This ﬁgure of 16.2% saying ‘yes’ in the interview compares with the 36.6% that indicated ‘yes’ in the forced- choice computer questionnaire earlier. In the interview, when asked how well they remembered having seen the video on a scale from 1 to 10, 37.0% (110 out of 297) indicated 1 (no memory at all) on the scale from 1 to 10, 63.0% indicated a 2 or above, and 15.8% indicated a score of 5 or above, with one participant (0.3%) indicating 10 (a very clear memory).

False memory versus false belief Similarly to the questionnaire, we categorized participants as Figure 2. Percentages of participants by experimental condition having an Interview False Memory when all the following who demonstrated a detailed Interview False Memory

Copyright © 2015 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 30:41–50 (2016) 46 L. Patihis and E. F. Loftus with prior suggestion scored marginally higher on the ques- somewhat familiar with the United 93 news story, compared tion ‘how well can you remember having seen the video on with those not familiar, were more likely to indicate a a scale from 1 to 10, where 1 means no memory at all and Persistent False Memory throughout the questionnaire and 10 means a very clear memory’ (no prior suggestion: M = interview (17.6% vs. 0.0%; χ2 (1, N = 202) = 12.06, exact 2.33, SD = 0.58; prior suggestion: M = 4.09, SD = p < .001, Cramer’s V = 0.244). 1.48), t(28) = 2.01, p = .054, Cohen’s d = 1.57. Levene’s test for equality of variances had a p-value of .15 (F = Media exposure on 9/11 2.24) indicating that assumption of equality of variance is A bivariate logistic regression showed that self-reported not strictly violated. However, the low Levene’s test news exposure on the day of 9/11 was a marginal predictor p-value and the small group size warrant reporting the equal of Persistent False Memory (Wald = 3.34, p = .068). Simi- variances not assumed test: comparing prior suggestion with larly, a bivariate logistic regression revealed that news expo- no prior suggestion on subsequent false memory clarity in the sure in the month after 9/11 was a predictor of Persistent interview yielded t(5.76) = 4.01, p = .008. False Memory (Wald = 4.89, p = .027). However, putting both these media exposure variables into the same model Memory Characteristics Questionnaire and Interview False resulted in neither being significant predictors of Persistent Memories False Memory (ps > .316), which may be explained by the Forty minutes before the interview, those 202 participants in high inter-correlation between the two media variables (r = the experimental condition answered MCQ items relating to .562, p < .001) although the variance inflation factor was the United 93 footage. Most of the items on the MCQ not excessive (VIF = 1.67). associated with the occurrence of an Interview False Memory (rs from .13 to .41; see Supporting Information Individual differences as correlates of Persistent False Table S1 for each MCQ item and effect sizes). The largest Memory MCQ predictor of Persistent False Memory was item 10 We use Persistent False Memory as the main outcome vari- ‘Feelings at the time that I first saw the video were’ (anchors: able in most of the analyses that follow because we consider 1=not intense,7=very intense), r = .41, p < .001. In other it the cleanest and most conservative measure of a real visual words, self-reported memory of more intense feelings was false memory. If participants indicated a false memory in the associated with a greater likelihood of reporting a detailed questionnaire but not the interview, it is possible that they false memory in the interview. did not actually visualize a false memory, perhaps misunder- standing the question. There is also some doubt cast on inter- Persistent False Memory view false memories if the participant had previously fi We de ned a Persistent False Memory as having occurred indicated that they had not seen the footage in the question- when a participant met the three criteria for a Questionnaire naire. Only two participants in the experimental group gave a False Memory and the three criteria for an Interview False false memory in the interview after not indicating a false Memory. By this metric, 25 (12.4%) participants exhibited memory in the questionnaire.3 Therefore, Persistent False a Persistent False Memory, out of the 202 who participated Memory is only a slightly more conservative measure than in both the suggestive questionnaire and interview. To clarify Interview False Memory (12.4% vs. 13.4% in experimental apparent differences in percentages, this Persistent False group). In addition, using Persistent False Memory elimi- fi fi Memory gure of 12% is by de nition from within the expe- nates the possible confound of some participants having rimental group only, whereas the aforementioned 10% Inter- prior suggestion and some not: all participants received prior view False Memory is from the whole sample (some of whom suggestion. For these reasons, and in order to reduce the ef- did not receive the prior suggestive questionnaire). fects of multiple comparisons that would occur if we used several outcome measures, we chose Persistent False Mem- Comparing those familiar to those not familiar with the ory as the best outcome measure for exploring most individ- news story ual differences measures. Although all participants were familiar with the 9/11 general news story, only 142 (70.3%) of the 202 in the experimental Demographics condition were somewhat familiar with the United 93 story There was no association between those who had a Persistent (responses to question ‘are you familiar with this event?’ False Memory and gender, age, or whether the participant > fi coded .5 = maybe, n =20;or1=yes, n = 122). In the question- was a psychology major or not (ps .332). Speci cally on naire, the 142 participants rated as being familiar with the gender, a variable discussed in previous research, females United 93 news story, compared with those 60 not familiar had statistically similar Persistent False Memory rates χ2 (coded 0 = no), were marginally more likely to indicate a (12.5%) to males (12.0%), (1, N = 202) = 0.009, p =.926 ’ Questionnaire False Memory (37.3% vs. 23.3%; χ2 (1, N = (exact p = 1.000), Cramer s V = 0.007. 202) = 6.50, p = .054, exact p = .072 (all ps two-tailed Those identifying their ethnicity as Middle Eastern had fi throughout), Cramer’s V = 0.136). In the interview, those signi cantly higher rates of Persistent False Memory somewhat familiar with the United 93 news story, compared 3 with those not familiar, were more likely to indicate an Inter- Because n = 2, we refrain from analyzing and discussing these individuals χ2 in depth in the article. We might speculate that in these individuals the initial view False Memory (19% vs. 0.0%; (1, N = 202) = 13.2, suggestion had a delayed effect, although further research is required to es- exact p < .001, Cramer’s V = 0.255). Similarly, those tablish this.

(46.7%) compared with those of other ethnicities (9.6%), χ2 gender, age, and socio-economic status of the interviewer (1, N = 202) = 17.57, exact p = .001, Cramer’s V = 0.295. also yielding no effect (cf. Porter, Birt, Yuille, & Lehman, 2000). This association between Middle Eastern ethnicity and Persistent False Memory was not affected by familiarity with Examining those ages 8–12 years at the time of the target the United 93 and was only slightly partially mediated by event (9/11) high-arousal negative emotion in the week following 9/11 (before mediation: β = 2.11, p < .001; after mediation: β = We replicated the aforementioned analysis excluding parti- 1.97, p = .001). cipants from age 23 to 51 years (n = 26) because the majority of our participants were between 18 and 22 years (n = 271), Reported emotion in week after 9/11 and, here, we wanted to focus on participants between the Those who demonstrated a Persistent False Memory re- ages of 18 and 22 years who were in mid-childhood (ages ported higher ratings of how often they had felt high-arousal 8 to 12 years) when the target events of 9/11 occurred. negative emotions in week after following 9/11 (M = 4.12, Almost all the comparisons had similar outcomes as with SD = 2.78), compared with those not exhibiting a Persistent the full sample. In other words, in those who were between False Memory (M = 3.14, SD = 2.12), t(197) = 1.66, p = 8 and 12 years old at the time of the target event (9/11), we .040, Cohen’s d = 0.40. This report of their emotions in the found similar percentages of false memories as with the full week after occurred approximately 10 years after 9/11. This dif- sample. We also confirmed that relationships between fami- ference was less pronounced in negative emotions not catego- liarity, ethnicity, alcohol use, fantasy proneness, and false rized as high arousal (ps > .133). memories existed in this 18- to 22-year-old subgroup, as well as the full sample. However, in the 18- to 22-year-old Alcohol use subgroup, we found all subtypes of memory for negative Those who demonstrated a Persistent False Memory re- emotions in the week following 9/11 predicted false memo- ported higher frequency of consuming alcohol than did those ries generally, whereas in the full sample, high-arousal who did not show a Persistent False Memory (M = 2.80: just emotions were a stronger predictor than other types of below the anchor twice a month, SD = 0.53), compared with emotion (see Supporting Information for full analysis). those not exhibiting a Persistent False Memory (M = 2.14, Figure 3 shows the percentages of Persistent False Memory just above the anchor once a month, SD = 1.39), t(200) = by age category at the time of the event. A chi-squared analysis 2.20, p = .029, Cohen’s d = 0.45. This Likert scale was a comparing Persistent False Memory rates of those who were fully anchored scale from 1 = never to 9 = everyday (from 12 years or younger (13.1%; consolidating first five age LaBrie, Hummer, Grant, & Luc, 2010). Similarly, those with categories in Figure 1) to those 13 years or older at the time a Persistent False Memory reported more heavy alcohol of the event (5.3%) revealed no significant difference, χ2 (1, drinking binges (defined as 4/5 drinks in a 2-hour period) N = 202) = 0.98, p =.323,Cramer’s V = 0.070. A chi-squared in the last 2 months (M = 2.76, SD = 2.05) compared with analysis on all the six categorized age groups revealed no main those who did not show a Persistent False Memory (M = effect for age group, χ2 (5, N = 202) = 4.83, p =.437,Cramer’s 1.94, SD = 1.59), t(200) = 2.33, p = .021, Cohen’s d = V = 0.155. This result was similarly not statistically significant 0.45. This Likert scale ranged from 1 = none,to9=10 or when excluding those age 23 years or above, χ2 (4, N =183)= more times. A summed composite of these and two other al- 3.66, p = .454, Cramer’s V = 0.141. Similarly, age analyzed as cohol use questions also revealed a statistically significant a continuous variable was not reliably associated with difference, t(200) = 2.01, p = .046, Cohen’s d = 0.28. Persistent False Memory (full age range: rbp = À0.069,

Fantasy proneness and dissociation Those with a Persistent False Memory were signiﬁcantly higher on fantasy proneness (M = 11.48, SD = 4.25; Creative Experiences Questionnaire) compared with those with no Persistent False Memory (M = 9.22, SD = 4.15), t(200) = 2.54, p = .012, Cohen’s d = 0.54. There were no differences between those with Persistent False Memory and those without on dissociation (p = .953; DES-C).

Personality We found no association between Persistent False Memory and participants’ personality trait subscales (Swedish Uni- versities Scale of Personality; see section on Method): social desirability, somatic trait anxiety, psychic trait anxiety, stress susceptibility, lack of assertiveness, impulsiveness, adventure seeking, detachment, embitterment, trait irritability, mistrust, verbal trait aggression, and physical trait aggression > Figure 3. Percentage of participants in each age category who (all ps .201). As described in the Supporting Information demonstrated a Persistent False Memory in the study. To convert we found no associations between the personality of the the x-axis to the age of the participants at the time of the study, add interviewer and false memory rates in the interview; with 10 years

Copyright © 2015 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 30:41–50 (2016) 48 L. Patihis and E. F. Loftus p = .332; limited 8–12 years age range at time of 9/11: previous indicated a false memory retracted, many claiming rbp =0.017,p =.819). that they had misunderstood the question. After debriefing, only 10% of the total sample in Smeets et al. (2009) indi- Qualitative analysis of interview false memories cated that they had had a false memory, compared with 66% before debriefing. In our current study, we also encoun- In the Supporting Information, we document some of the tered a drop in the number of false memories when we wording used by participants in the interview who demon- explicitly clarified the events and the question in our face- strated an Interview False Memory. In those two examples, to-face interview (from 33% to 13% within the experimental it is clear that the participants understood which crash the in- group). Our study shows that this reduction in false memory terviewer was referring to. The United 93 crash was clearly rates can be achieved before debriefing, which is to say that distinguished from the other 9/11 crashes at the beginning the reduction in our study was not caused by the post-event of the interview (see the full interview script in the true information (as could have been a partial cause of the Supporting Information). In both interview excerpts, the reduction in Smeets et al., 2009). The reduction in false emerging false memory seems relatively fragile at this stage, memory rates in the current study is perhaps because the and although some false details are emerging (e.g., ‘burst fl fl ’ interview was clearer, more cautious, and more detailed into ames when it crashed into the oor ), it is also clear compared with the briefer questionnaire. One could also that these details are not yet filled out into a vivid and heavily speculate that the face-to-face nature of the interview in our detailed memory. This pattern was observed in many of the study elicited more caution and attention in the participants, interview transcripts that demonstrated a false memory. It compared with the anonymous clicking of the mouse on the is because of this qualitative examination of the words used computer questionnaire (cf. Koriat & Goldsmith, 1996). by participants in the interview that we conclude that the From this, we concur with Smeets et al. (2009) that even false memories are still in the fragile early stages of forma- though there is a reduction of false memory rates, there is tion and not yet fully developed into multi-detailed and vivid still a minority of people who exhibit what appears to be a autobiographical memory. genuine false memory. It is clear from both these studies that a good proportion of the emerging false memories are ini- DISCUSSION tially fragile and labile, because many initial false reports are corrected in the presence of true post-event information The study produced false memory reports of seeing the non- or more cautious questioning. However, it is possible that existent footage of the United 93 crash in a sizable minority even some of these labile false reports would consolidate of subjects in both in the computer questionnaire and in our into full and stable false memories in the presence of rein- more rigorous follow-up interview. This interview involved forcement and time and in the absence of debriefing or cau- free recall (as opposed to forced-choice in the questionnaire), tion. Indeed, in the present study, we show that two expo- and we asked follow-up questions to help us distinguish sures to misleading information led to higher false memory between a false memory and a false belief. We were conser- rates than just one. vative in that we required three prerequisite necessary condi- On a related note, Ost et al. (2008) and Smeets et al. tions (a ‘yes’, a score not including 1 = no memory at all on a (2009) raised the question whether false memories observed 1–10 scale, and a false detail of the actual crash) when in crashing memory studies were already formed prior to distinguishing a false memory from a false belief in both suggestion (spontaneous) or caused by the suggestion in the questionnaire and the interview. Even with the necessary the experiment itself. Our use of random assignment during condition of a false detail accompanying the false report, we the questionnaire phase into groups that either received sug- still found 12.4% had a Persistent False Memory throughout gestion or no suggestion allowed us to conclude that sugges- the questionnaire and the interview. Of those familiar with tion in the experimental condition is an important cause of the United 93 news story, 17.6% had a Persistent False the false memories. This is evidence that perhaps a good pro- Memory, with media exposure marginally associating with portion of false memory reports in crashing memory studies such false memories. Prior suggestion 40 minutes earlier in are in fact suggestion-dependent rather than spontaneous. the questionnaire had an additive effect on false memory Qualitative evidence for this fragility of these newly rates in the interview, with approximately a fourfold differ- formed false memories is perhaps gained from examining ence in percentages. Middle Eastern participants had higher the actual wording used by the participants in the interview. rates of Persistent False Memory than people of other ethnic- It appears that even in those demonstrating false memories, ities. High-arousal negative emotion, alcohol use, and those false memories are newly formed with only a few fantasy proneness were associated with higher Persistent details. They appear relatively labile, and this fragility may False Memory. explain why previous research found that many false reports Despite many previous studies showing co-occurrence of were retracted after correcting post-event information a memory report and a false detail, some uncertainty about (Smeets et al., 2009) and why our percentages dropped from the paradigm was introduced by Smeets et al. (2009). That the questionnaire to the interview. It is for these reasons that study raised the question of whether the crashing memory we suspect that repeated reinforcement and time would be paradigm was actually producing false memories, as needed to solidify the false memories into elaborate and opposed to mistaken beliefs or other types of errors. In highly detailed accounts. This is supported by our finding Smeets et al. (2009), they found that after participants were that prior suggestion seemed to also lead to clearer false told there was actually no footage, 80% of those who had memories, compared with those with no prior suggestion.

In addition, past research into implanting childhood false could attempt that. In addition, this study is different than memories has shown that repetition (e.g., Ceci, Huffman, many previous studies in that a sizable minority of partici- Smith, & Loftus, 1994a) and repetition combined with sug- pants was not familiar with the specific United 93 news gestion (e.g., Ceci, Loftus, Leichtman, & Bruck, 1994b) story. Nevertheless, this allowed us to confirm that familiar- might be important for producing elaborate false memories ity was indeed related to false memory rates. that become richer in detail and more resistant to correction, This current study contributes to our current knowledge of compared with the early stages of false memory formation. false memory and crashing memory research in a number of As in previous crashing memory studies (Ost et al., 2008), ways. Firstly, it shows that reductions in false memory rates we found that fantasy proneness was associated with false can be found by being more cautious and detailed in the way memory. This can be explained by the ability of fantasy you ask the crashing memory questions, even after mislead- prone individuals to visualize and imagine events and a ing suggestion has been presented. The current study estab- tendency to remember a lot of visual details from the past lishes such a reduction in rates before debriefing, whereas (both accurately and inaccurately, see Merckelbach, 2004; Smeets et al. (2009) did so after also telling the participants cf. Patihis et al., 2013). This capacity likely helps bring the that no footage actually exists. It also demonstrates that not event to mind in a visual form, which in turn leads to the all false memories in the crashing memory paradigm are so later consolidation of a full false memory. However, unlike fragile that they either disappear with cautious further probing, Ost et al. (2008), we did not find that dissociation was asso- or turn out to be a mere belief. We establish here that a mino- ciated with false memory production. Other personality traits rity are persistent and genuine detailed false memories. We also did not associate with false memories in the current argue that in the absence of debriefing that these semi- study. In particular, it is interesting that those who were high autobiographical false memories could persist, especially if on social desirability had similar rates of false memory as reinforced. We show that repeated suggestion increased false those low on social desirability. This tends to dampen the memory rates in the crashing memory paradigm. We also iden- idea that participants may be demonstrating false memories tify factors that might increase these types of autobiographical out of a wish to be socially accepted by the researchers. memories, such as high-arousal emotions, alcohol use, prior Interestingly, we found that participants identifying as suggestion, and familiarity. We also present some evidence Middle Eastern had higher false memory rates, compared that is congruent with previous findings showing a link with other ethnicities. It is unclear why this might be. One between fantasy proneness and false memories. This study, possibility was that because of the political connection of and the crashing memory paradigm as a whole, gives memory the terrorist attacks on 9/11 to the Middle East, those identi- researchers the opportunity to measure memory distortion in fying as Middle Eastern might be more familiar with the the autobiographical domain with the rare knowledge that we news story. This, however, is probably not the case because are sure the original event did not take place. News events such those identifying as Middle Eastern had similar familiarity as 9/11 are important because they represent memories that with the United 93 news event as other participants. Another can subjectively be experienced as flashbulb-like. Such collec- possibility is that they had more negative emotions during tive experience represents the part of our autobiographical and following 9/11: but this idea had only marginal support: memory that is shared with millions of others. As Neisser memory of negative emotion only partially explained the (1982, p. 48) wrote, these are events in which we ‘line up relationship in a mediation analysis. For these reasons, our own lives with the course of history itself and say “Iwas although this result seems intuitively to make sense, it is there”’. It could be considered rather Orwellian that these difficult to say for certain the reason for the relationship news memories are malleable. between ethnicity and false memory for 9/11. In addition, caution is warranted here because only a small sample of fi 22 identi ed as Middle Eastern. ACKNOWLEDGEMENTS Interestingly, those who reported remembering more fre- quent experience of high-arousal negative emotions in the We thank Steven Frenda for his help with the design and data week following 11 September 2001, had a higher proportion collection stages of the study and research assistants Marlene of false memories for the United 93 crash footage. This may Lin Ma, Anita Chen, Joseph McCall, Matthew Miller, be an indication that high emotional arousal may not only aid Lavina Ho, Stephany Debski, Stephanie M. Martinez, in memory consolidation (Cahill & McGaugh, 1998) but Zahra Bozorgkhan, Kenneth Nguyen, Patricia Place, paradoxically can lead to errors in details, in particular a Gina Machiaverna, Alexandra Chindris, Samuel Cretcher, higher susceptibility to false memories for some details (cf. Monica Aguilar, Paddy Asgari, Yesenia Orozco, Luis Garcia, Christianson & Loftus, 1987; also see the Paradoxical Nega- Natalie Ross, Devan Duenas, Howard Lee, and Hillary Patton tive Emotion hypothesis, Porter, Taylor, & ten Brinke, 2008). for their help during data collection. There are some limitations to this study. Our measure for how often participants felt negative emotions in the week after 9/11 would not be as accurate as if we had asked them REFERENCES soon after the event. This possible inaccuracy in this variable may come from the passage of time, or changing of how they Cahill, L., & McGaugh, J. L. (1998). Mechanisms of emotional arousal and lasting declarative memory. Trends in Neurosciences, 21, 294–299. appraise the events of 9/11 (cf. Levine, Whalen, Henker, & Ceci, S. J., Huffman, M. L. C., Smith, E., & Loftus, E. F. (1994a). Repeat- Jamner, 2005). One method for addressing this issue would edly thinking about a non-event: Source misattributions among pre- be to conduct a longitudinal design, and future research schoolers. Consciousness and Cognition, 3, 388–407.

Ceci, S. J., Loftus, E. F., Leichtman, M. D., & Bruck, M. (1994b). The Merckelbach, H. (2004). Telling a good story: Fantasy proneness and the possible role of source misattributions in the creation of false beliefs quality of fabricated memories. Personality & Individual Differences, among preschoolers. International Journal of Clinical and Experimental 37, 1371–1382. Hypnosis, 42, 304–320. Merckelbach, H., Horselenberg, R., & Muris, P. (2001). The Creative Expe- Christianson, S. Å., & Loftus, E. F. (1987). Memory for traumatic events. riences Questionnaire (CEQ): A brief self-report measure of fantasy Applied Cognitive Psychology, 1, 225–239. proneness. Personality and Individual Differences, 31, 987–995. Crombag, H. F. M., Wagenaar, W. A., & van Koppen, P. J. (1996). Crashing Merckelbach, H., Muris, P., & Rassin, E. (1999). Fantasy proneness and memories and the problem of ‘source monitoring’. Applied Cognitive cognitive failures as correlates of dissociative experiences. Personality Psychology, 10,95–104. and Individual Differences, 26, 961–967. Garfinkel, S. N., Dienes, Z., & Duka, T. (2006). The effect of alcohol and Neisser, U. (1982). Snapshots or benchmarks. In U. Neisser (Ed.), Memory repetition at encoding on implicit and explicit false memories. Psycho- observed: Remembering in natural contexts (edn, pp. 43–48). San pharmacology, 188, 498–508. Francisco, CA: Freeman. Garry, M., Manning, C. G., Loftus, E. F., & Sherman, S. J. (1996). Imagina- Ost, J., Granhag, P., Udell, J., & Hjelmsäter, E. R. (2008). Familiarity breeds tion inflation: Imagining a childhood event inflates confidence that it distortion: The effects of media exposure on false reports concerning occurred. Psychonomic Bulletin and Review, 3, 208–214. media coverage of the terrorist attacks in London on 7 July 2005. Granhag, P. A., Stromwall, L., & Billings, J. (2003). I’ll never forget the Memory, 16,76–85. sinking ferry: How social influence makes false memories surface. In Ost, J., Hogbin, I., & Granhag, P. (2006). Altering false reports via confed- M. Vanderhallen, G. Vervaeke, P. J. van Koppen, & J. Goethals (Eds.), erate influence. Social Influence, 1, 105–116. Much ado about crime: Chapters on psychology and law (pp. 129–140). Ost, J., Vrij, A., Costall, A., & Bull, R. (2002). Crashing memories and re- Politeia: Brussels. ality monitoring: Distinguishing between perceptions, imaginings and Gustavsson, J. P., Bergman, H., Edman, G., Ekselius, L., Von Knorring, L., false memories. Applied Cognitive Psychology, 16, 125–134. & Linder, J. (2000). Swedish universities Scales of Personality (SSP): Patihis, L. (2012). Memory distortions: Individual differences and paradigm Construction, internal consistency and normative data. Acta Psychiatrica comparisons (Master’s thesis). University of California, Irvine. UMI Scandinavica, 102, 217–225. 1512095. Jelicic, M., Smeets, T., Candel, I., van Suijdam, M., & Merckelbach, H. Patihis, L., Frenda, S. J., LePort, A. K., Petersen, N., Nichols, R. M., Stark, (2006a). No, I don’t remember seeing video footage of the killing of Theo C. E., … Loftus, E. F. (2013). False memories in highly superior autobio- van Gogh! Misinformation manipulations do not always elicit false graphical memory individuals. Proceedings of the National Academy of memories. In K. Nixon (Ed.), Forensic recall and eyewitness testimony Sciences, 110, 20947–20952. (pp. 21–25). London, UK: IA-IP Publishing. Patihis, L., & Younes Burton, H. J. (in press). False memories in therapy and Jelicic, M., Smeets, T., Peters, M. J. V., Candel, I., Horselenberg, R., & hypnosis before 1980. Psychology of Consciousness: Theory, Research, Merckelbach, H. (2006b). Assassination of a controversial politician: and Practice. Remembering details from another non-existent film. Applied Cognitive Porter, S., Birt, A. R., Yuille, J. C., & Lehman, D. R. (2000). Negotiating Psychology, 20, 591–596. false memories: Interviewer and rememberer characteristics relate to Johnson, M. K., Foley, M. A., Suengas, A. G., & Raye, C. L. (1988). memory distortion. Psychological Science, 11, 507–510. Phenomenal characteristics of memories for perceived and imagined Porter, S., Taylor, K., & ten Brinke, L. (2008). Memory for media: Investi- autobiographical events. Journal of Experimental Psychology. General, gation of false memories for negatively and positively charged public 117, 371–376. events. Memory, 16, 658–666. Koriat, A., & Goldsmith, M. (1996). Monitoring and control processes in the Roediger, H. L., & McDermott, K. B. (1995). Creating false memories: Re- strategic regulation of memory accuracy. Psychological Review, 103, membering words not presented in lists. Journal of Experimental Psy- 490–517. chology: Learning, Memory, and Cognition, 21, 803–814. LaBrie, J. W., Hummer, J. F., Grant, S., & Lac, A. (2010). Immediate reduc- Smeets, T., Jelicic, M., Peters, M. J. V., Candel, I., Horselenberg, R., & tions in misperceived social norms among high-risk college student Merckelbach, H. (2006). Of course I remember seeing that film! How groups. Addictive Behaviors, 35, 1094–1101. ambiguous questions generate crashing memories. Applied Cognitive Levine, L. J., Whalen, C. K., Henker, B., & Jamner, L. D. (2005). Looking Psychology, 20, 779–789. back on September 11, 2001: Appraised impact and memory for emotions Smeets, T., Telgen, S., Ost, J., Jelicic, M., & Merckelbach, H. (2009). in adolescents and adults. Journal of Adolescent Research, 20, 497–523. What’s behind crashing memories? Plausibility, belief, and memory in Loftus, E. F., & Pickrell, J. E. (1995). The formation of false memories. reports of having seen non-existent images. Applied Cognitive Psychology, Psychiatric Annals, 25, 720–725. 23, 1333–1341. Loftus, E. F., Miller, D. G., & Burns, H. J. (1978). Semantic integration of Wilson, K., & French, C. C. (2006). The relationship between susceptibility verbal information into a visual memory. Journal of Experimental to false memories, dissociativity, and paranormal belief and experience. Psychology, 4,19–31. Personality and Individual Differences, 41, 1493–1502. Lynn, S. J., Lilienfeld, S. O., Merckelbach, H., Giesbrecht, T., McNally, R. Wright, D. B., & Loftus, E. F. (1999). Measuring dissociation: Comparison J., Loftus, E. F., … & Malaktaris, A. (2014). The trauma model of disso- of alternative forms of the dissociative experiences scale. American Journal ciation: Inconvenient truths and stubborn fictions. Comment on of Psychology, 112,497–519. Dalenberg et al. (2012). Psychological Bulletin, 140, 896–910. Mazzoni, G. (2002). Naturally occurring and suggestion-dependent memory distortions: The convergence of disparate research traditions. European Psychologist, 7,17–30. SUPPORTING INFORMATION McCloskey, M., & Zaragoza, M. (1985). Misleading postevent information Supporting information may be found in the online version of this and memory for events: Arguments and evidence against memory impair- article. ment hypotheses. Journal of Experimental Psychology: General, 114,1–16.